Can't Optimize or Compile YoloV8 with trained on Custom Dataset

General
1

I trained a model with yolov8s and yolov8n using the instructions and dockerfile from here: hailo_model_zoo/training/yolov8 at master · hailo-ai/hailo_model_zoo · GitHub

In Yolo Docker:

yolo detect train data=/app/yolotrain.yaml model=yolov8s epochs=100 batch=16 imgsz=446 name=8s_v1.0

yolo detect train data=/app/yolotrain.yaml model=yolov8n epochs=100 batch=16 imgsz=446 name=8n_v1.0

yolo export model=/workspace/ultralytics/runs/detect/8n_v1.0/weights/best.pt imgsz=448 format=onnx opset=11

yolo export model=/workspace/ultralytics/runs/detect/8s_v1.0/weights/best.pt imgsz=448 format=onnx opset=11

I then copied those to my hailo_ai_sw_2024-07.1_container so that I can compile (or parse, optimize, then compile, I’m still a little foggy on why some tutorials skip straight to compile and others parse and optimize)

I updated my yolo.yaml to account for the new input size:

base:
- base/coco.yaml
preprocessing:
  network_type: detection
  input_shape:
  - 448
  - 448
  - 3
  meta_arch: yolo_v5
  padding_color: 114
postprocessing:
  nms_iou_thresh: 0.45
  score_threshold: 0.01
  meta_arch: yolo_v5
  anchors:
    strides:
    - 8
    - 16
    - 32
    sizes:
    - - 10
      - 13
      - 16
      - 30
      - 33
      - 23
    - - 30
      - 61
      - 62
      - 45
      - 59
      - 119
    - - 116
      - 90
      - 156
      - 198
      - 373
      - 326
  hpp: false
info:
  source: https://github.com/ultralytics/yolov5/releases/tag/v2.0
parser:
  normalization_params:
    normalize_in_net: true
    mean_list:
    - 0.0
    - 0.0
    - 0.0
    std_list:
    - 255.0
    - 255.0
    - 255.0
evaluation:
  labels_offset: 1
  classes: 80
  dataset_name: coco_2017_detection

image
image1342×763 31.9 KB

Parsing works properly:

hailomz parse --ckpt 8s_v1.0_hel.onnx --hw-arch hailo8l yolov8s

<Hailo Model Zoo INFO> Start run for network yolov8s ...
<Hailo Model Zoo INFO> Initializing the runner...
[info] Translation started on ONNX model yolov8s
[info] Restored ONNX model yolov8s (completion time: 00:00:00.24)
[info] Extracted ONNXRuntime meta-data for Hailo model (completion time: 00:00:00.86)
[info] NMS structure of yolov8 (or equivalent architecture) was detected.
[info] In order to use HailoRT post-processing capabilities, these end node names should be used: /model.22/cv2.0/cv2.0.2/Conv /model.22/cv3.0/cv3.0.2/Conv /model.22/cv2.1/cv2.1.2/Conv /model.22/cv3.1/cv3.1.2/Conv /model.22/cv2.2/cv2.2.2/Conv /model.22/cv3.2/cv3.2.2/Conv.
[info] Start nodes mapped from original model: 'images': 'yolov8s/input_layer1'.
[info] End nodes mapped from original model: '/model.22/cv2.0/cv2.0.2/Conv', '/model.22/cv3.0/cv3.0.2/Conv', '/model.22/cv2.1/cv2.1.2/Conv', '/model.22/cv3.1/cv3.1.2/Conv', '/model.22/cv2.2/cv2.2.2/Conv', '/model.22/cv3.2/cv3.2.2/Conv'.
[info] Translation completed on ONNX model yolov8s (completion time: 00:00:01.56)
[info] Saved HAR to: /jtech/yolov8s.har

and

hailomz parse --ckp 8n_v1.0_hel.onnx --hw-arch hailo8l yolov8n

<Hailo Model Zoo INFO> Start run for network yolov8n ...
<Hailo Model Zoo INFO> Initializing the runner...
[info] Translation started on ONNX model yolov8n
[info] Restored ONNX model yolov8n (completion time: 00:00:00.04)
[info] Extracted ONNXRuntime meta-data for Hailo model (completion time: 00:00:00.24)
[info] NMS structure of yolov8 (or equivalent architecture) was detected.
[info] In order to use HailoRT post-processing capabilities, these end node names should be used: /model.22/cv2.0/cv2.0.2/Conv /model.22/cv3.0/cv3.0.2/Conv /model.22/cv2.1/cv2.1.2/Conv /model.22/cv3.1/cv3.1.2/Conv /model.22/cv2.2/cv2.2.2/Conv /model.22/cv3.2/cv3.2.2/Conv.
[info] Start nodes mapped from original model: 'images': 'yolov8n/input_layer1'.
[info] End nodes mapped from original model: '/model.22/cv2.0/cv2.0.2/Conv', '/model.22/cv3.0/cv3.0.2/Conv', '/model.22/cv2.1/cv2.1.2/Conv', '/model.22/cv3.1/cv3.1.2/Conv', '/model.22/cv2.2/cv2.2.2/Conv', '/model.22/cv3.2/cv3.2.2/Conv'.
[info] Translation completed on ONNX model yolov8n (completion time: 00:00:00.81)
[info] Saved HAR to: /jtech/yolov8n.har

but I get the same error if I try and compile or optimize:

hailomz compile yolov8s --hw-arch hailo8l --har ./yolov8s.har --calib-path ./training/yolo/images/

<Hailo Model Zoo INFO> Start run for network yolov8s ...
<Hailo Model Zoo INFO> Initializing the runner...
[info] Translation started on ONNX model yolov8s
[info] Restored ONNX model yolov8s (completion time: 00:00:00.15)
[info] Extracted ONNXRuntime meta-data for Hailo model (completion time: 00:00:00.65)
[info] NMS structure of yolov8 (or equivalent architecture) was detected.
[info] In order to use HailoRT post-processing capabilities, these end node names should be used: /model.22/cv2.0/cv2.0.2/Conv /model.22/cv3.0/cv3.0.2/Conv /model.22/cv2.1/cv2.1.2/Conv /model.22/cv3.1/cv3.1.2/Conv /model.22/cv2.2/cv2.2.2/Conv /model.22/cv3.2/cv3.2.2/Conv.
[info] Start nodes mapped from original model: 'images': 'yolov8s/input_layer1'.
[info] End nodes mapped from original model: '/model.22/cv2.0/cv2.0.2/Conv', '/model.22/cv3.0/cv3.0.2/Conv', '/model.22/cv2.1/cv2.1.2/Conv', '/model.22/cv3.1/cv3.1.2/Conv', '/model.22/cv2.2/cv2.2.2/Conv', '/model.22/cv3.2/cv3.2.2/Conv'.
[info] Translation completed on ONNX model yolov8s (completion time: 00:00:01.38)
[info] Saved HAR to: /jtech/yolov8s.har
(hailo_virtualenv) hailo@docker-desktop:/jtech$ hailomz compile yolov8s --hw-arch hailo8l --har ./yolov8s.har --calib-path ./training/yolo/images/
<Hailo Model Zoo INFO> Start run for network yolov8s ...
<Hailo Model Zoo INFO> Initializing the hailo8l runner...
<Hailo Model Zoo INFO> Preparing calibration data...
[info] Loading model script commands to yolov8s from /local/workspace/hailo_model_zoo/hailo_model_zoo/cfg/alls/generic/yolov8s.alls
[info] Starting Model Optimization
[warning] Reducing optimization level to 0 (the accuracy won't be optimized and compression won't be used) because there's no available GPU
[warning] Running model optimization with zero level of optimization is not recommended for production use and might lead to suboptimal accuracy results
[info] Model received quantization params from the hn
Traceback (most recent call last):
  File "/local/workspace/hailo_virtualenv/bin/hailomz", line 33, in <module>
    sys.exit(load_entry_point('hailo-model-zoo', 'console_scripts', 'hailomz')())
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main.py", line 122, in main
    run(args)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main.py", line 111, in run
    return handlers[args.command](args)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main_driver.py", line 250, in compile
    _ensure_optimized(runner, logger, args, network_info)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main_driver.py", line 91, in _ensure_optimized
    optimize_model(
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/core/main_utils.py", line 321, in optimize_model
    runner.optimize(calib_feed_callback)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_common/states/states.py", line 16, in wrapped_func
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/runner/client_runner.py", line 2093, in optimize
    self._optimize(calib_data, data_type=data_type, work_dir=work_dir)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_common/states/states.py", line 16, in wrapped_func
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/runner/client_runner.py", line 1935, in _optimize
    self._sdk_backend.full_quantization(calib_data, data_type=data_type, work_dir=work_dir)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/sdk_backend/sdk_backend.py", line 1045, in full_quantization
    self._full_acceleras_run(self.calibration_data, data_type)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/sdk_backend/sdk_backend.py", line 1229, in _full_acceleras_run
    optimization_flow.run()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 306, in wrapper
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/flows/optimization_flow.py", line 316, in run
    step_func()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 250, in wrapped
    result = method(*args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/subprocess_wrapper.py", line 121, in parent_wrapper
    self.build_model()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 250, in wrapped
    result = method(*args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/flows/optimization_flow.py", line 230, in build_model
    model.compute_output_shape(shapes)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/model/hailo_model/hailo_model.py", line 1013, in compute_output_shape
    return self.compute_and_verify_output_shape(input_shape, verify_layer_inputs_shape=False)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/model/hailo_model/hailo_model.py", line 1047, in compute_and_verify_output_shape
    layer_output_shape = layer.compute_output_shape(layer_input_shapes)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/keras/engine/base_layer.py", line 917, in compute_output_shape
    outputs = self(inputs, training=False)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "/tmp/__autograph_generated_filevv_9ae60.py", line 41, in tf__call
    outputs = ag__.converted_call(ag__.ld(self).call_core, (ag__.ld(inputs), ag__.ld(training)), dict(**ag__.ld(kwargs)), fscope)
  File "/tmp/__autograph_generated_file5sv35h47.py", line 90, in tf__call_core
    ag__.if_stmt((ag__.ld(self).postprocess_type in [ag__.ld(PostprocessType).NMS, ag__.ld(PostprocessType).BBOX_DECODER]), if_body_3, else_body_3, get_state_3, set_state_3, ('do_return', 'retval_'), 2)
  File "/tmp/__autograph_generated_file5sv35h47.py", line 22, in if_body_3
    retval_ = ag__.converted_call(ag__.ld(self).bbox_decoding_and_nms_call, (ag__.ld(inputs),), dict(is_bbox_decoding_only=(ag__.ld(self).postprocess_type == ag__.ld(PostprocessType).BBOX_DECODER)), fscope)
  File "/tmp/__autograph_generated_filexgc1j34i.py", line 99, in tf__bbox_decoding_and_nms_call
    ag__.if_stmt((ag__.ld(self).meta_arch in [ag__.ld(NMSOnCpuMetaArchitectures).YOLOV5, ag__.ld(NMSOnCpuMetaArchitectures).YOLOX]), if_body_4, else_body_4, get_state_4, set_state_4, ('decoded_bboxes', 'detection_score', 'do_return', 'retval_', 'inputs'), 4)
  File "/tmp/__autograph_generated_filexgc1j34i.py", line 96, in else_body_4
    ag__.if_stmt((ag__.ld(self).meta_arch == ag__.ld(NMSOnCpuMetaArchitectures).YOLOV5_SEG), if_body_3, else_body_3, get_state_3, set_state_3, ('decoded_bboxes', 'detection_score', 'do_return', 'retval_'), 4)
  File "/tmp/__autograph_generated_filexgc1j34i.py", line 93, in else_body_3
    ag__.if_stmt((ag__.ld(self).meta_arch == ag__.ld(NMSOnCpuMetaArchitectures).YOLOV8), if_body_2, else_body_2, get_state_2, set_state_2, ('decoded_bboxes', 'detection_score'), 2)
  File "/tmp/__autograph_generated_filexgc1j34i.py", line 69, in if_body_2
    (decoded_bboxes, detection_score) = ag__.converted_call(ag__.ld(self).yolov8_decoding_call, (ag__.ld(inputs),), None, fscope)
  File "/tmp/__autograph_generated_file7isggz0k.py", line 67, in tf__yolov8_decoding_call
    decoded_bboxes = ag__.converted_call(ag__.ld(tf).expand_dims, (ag__.ld(decoded_bboxes),), dict(axis=2), fscope)
ValueError: Exception encountered when calling layer "yolov8_nms_postprocess" (type HailoPostprocess).

in user code:

    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/base_hailo_none_nn_core_layer.py", line 43, in call  *
        outputs = self.call_core(inputs, training, **kwargs)
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 113, in call_core  *
        inputs,
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 148, in bbox_decoding_and_nms_call  *
        decoded_bboxes, detection_score = self.yolov8_decoding_call(inputs)
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 355, in yolov8_decoding_call  *
        decoded_bboxes = tf.expand_dims(decoded_bboxes, axis=2)

    ValueError: Tried to convert 'input' to a tensor and failed. Error: None values not supported.


Call arguments received by layer "yolov8_nms_postprocess" (type HailoPostprocess):
  • inputs=['tf.Tensor(shape=(None, 56, 56, 64), dtype=float32)', 'tf.Tensor(shape=(None, 56, 56, 2), dtype=float32)', 'tf.Tensor(shape=(None, 28, 28, 64), dtype=float32)', 'tf.Tensor(shape=(None, 28, 28, 2), dtype=float32)', 'tf.Tensor(shape=(None, 14, 14, 64), dtype=float32)', 'tf.Tensor(shape=(None, 14, 14, 2), dtype=float32)']
  • training=False
  • kwargs=<class 'inspect._empty'>

and for 8n:

hailomz compile yolov8n --hw-arch hailo8l --har ./yolov8n.har --calib-pa
th ./training/yolo/images/

<Hailo Model Zoo INFO> Start run for network yolov8n ...
<Hailo Model Zoo INFO> Initializing the hailo8l runner...
<Hailo Model Zoo INFO> Preparing calibration data...
[info] Loading model script commands to yolov8n from /local/workspace/hailo_model_zoo/hailo_model_zoo/cfg/alls/generic/yolov8n.alls
[info] Starting Model Optimization
[warning] Reducing optimization level to 0 (the accuracy won't be optimized and compression won't be used) because there's no available GPU
[warning] Running model optimization with zero level of optimization is not recommended for production use and might lead to suboptimal accuracy results
[info] Model received quantization params from the hn
Traceback (most recent call last):
  File "/local/workspace/hailo_virtualenv/bin/hailomz", line 33, in <module>
    sys.exit(load_entry_point('hailo-model-zoo', 'console_scripts', 'hailomz')())
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main.py", line 122, in main
    run(args)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main.py", line 111, in run
    return handlers[args.command](args)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main_driver.py", line 250, in compile
    _ensure_optimized(runner, logger, args, network_info)
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/main_driver.py", line 91, in _ensure_optimized
    optimize_model(
  File "/local/workspace/hailo_model_zoo/hailo_model_zoo/core/main_utils.py", line 321, in optimize_model
    runner.optimize(calib_feed_callback)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_common/states/states.py", line 16, in wrapped_func
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/runner/client_runner.py", line 2093, in optimize
    self._optimize(calib_data, data_type=data_type, work_dir=work_dir)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_common/states/states.py", line 16, in wrapped_func
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/runner/client_runner.py", line 1935, in _optimize
    self._sdk_backend.full_quantization(calib_data, data_type=data_type, work_dir=work_dir)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/sdk_backend/sdk_backend.py", line 1045, in full_quantization
    self._full_acceleras_run(self.calibration_data, data_type)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_sdk_client/sdk_backend/sdk_backend.py", line 1229, in _full_acceleras_run
    optimization_flow.run()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 306, in wrapper
    return func(self, *args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/flows/optimization_flow.py", line 316, in run
    step_func()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 250, in wrapped
    result = method(*args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/subprocess_wrapper.py", line 121, in parent_wrapper
    self.build_model()
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/tools/orchestator.py", line 250, in wrapped
    result = method(*args, **kwargs)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/flows/optimization_flow.py", line 230, in build_model
    model.compute_output_shape(shapes)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/model/hailo_model/hailo_model.py", line 1013, in compute_output_shape
    return self.compute_and_verify_output_shape(input_shape, verify_layer_inputs_shape=False)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/model/hailo_model/hailo_model.py", line 1047, in compute_and_verify_output_shape
    layer_output_shape = layer.compute_output_shape(layer_input_shapes)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/keras/engine/base_layer.py", line 917, in compute_output_shape
    outputs = self(inputs, training=False)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "/tmp/__autograph_generated_filezehnp_ey.py", line 41, in tf__call
    outputs = ag__.converted_call(ag__.ld(self).call_core, (ag__.ld(inputs), ag__.ld(training)), dict(**ag__.ld(kwargs)), fscope)
  File "/tmp/__autograph_generated_filezg3153uh.py", line 90, in tf__call_core
    ag__.if_stmt((ag__.ld(self).postprocess_type in [ag__.ld(PostprocessType).NMS, ag__.ld(PostprocessType).BBOX_DECODER]), if_body_3, else_body_3, get_state_3, set_state_3, ('do_return', 'retval_'), 2)
  File "/tmp/__autograph_generated_filezg3153uh.py", line 22, in if_body_3
    retval_ = ag__.converted_call(ag__.ld(self).bbox_decoding_and_nms_call, (ag__.ld(inputs),), dict(is_bbox_decoding_only=(ag__.ld(self).postprocess_type == ag__.ld(PostprocessType).BBOX_DECODER)), fscope)
  File "/tmp/__autograph_generated_filen0w1t1lj.py", line 99, in tf__bbox_decoding_and_nms_call
    ag__.if_stmt((ag__.ld(self).meta_arch in [ag__.ld(NMSOnCpuMetaArchitectures).YOLOV5, ag__.ld(NMSOnCpuMetaArchitectures).YOLOX]), if_body_4, else_body_4, get_state_4, set_state_4, ('decoded_bboxes', 'detection_score', 'do_return', 'retval_', 'inputs'), 4)
  File "/tmp/__autograph_generated_filen0w1t1lj.py", line 96, in else_body_4
    ag__.if_stmt((ag__.ld(self).meta_arch == ag__.ld(NMSOnCpuMetaArchitectures).YOLOV5_SEG), if_body_3, else_body_3, get_state_3, set_state_3, ('decoded_bboxes', 'detection_score', 'do_return', 'retval_'), 4)
  File "/tmp/__autograph_generated_filen0w1t1lj.py", line 93, in else_body_3
    ag__.if_stmt((ag__.ld(self).meta_arch == ag__.ld(NMSOnCpuMetaArchitectures).YOLOV8), if_body_2, else_body_2, get_state_2, set_state_2, ('decoded_bboxes', 'detection_score'), 2)
  File "/tmp/__autograph_generated_filen0w1t1lj.py", line 69, in if_body_2
    (decoded_bboxes, detection_score) = ag__.converted_call(ag__.ld(self).yolov8_decoding_call, (ag__.ld(inputs),), None, fscope)
  File "/tmp/__autograph_generated_filepm3y5f6c.py", line 67, in tf__yolov8_decoding_call
    decoded_bboxes = ag__.converted_call(ag__.ld(tf).expand_dims, (ag__.ld(decoded_bboxes),), dict(axis=2), fscope)
ValueError: Exception encountered when calling layer "yolov8_nms_postprocess" (type HailoPostprocess).

in user code:

    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/base_hailo_none_nn_core_layer.py", line 43, in call  *
        outputs = self.call_core(inputs, training, **kwargs)
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 113, in call_core  *
        inputs,
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 148, in bbox_decoding_and_nms_call  *
        decoded_bboxes, detection_score = self.yolov8_decoding_call(inputs)
    File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/acceleras/hailo_layers/hailo_postprocess.py", line 355, in yolov8_decoding_call  *
        decoded_bboxes = tf.expand_dims(decoded_bboxes, axis=2)

    ValueError: Tried to convert 'input' to a tensor and failed. Error: None values not supported.


Call arguments received by layer "yolov8_nms_postprocess" (type HailoPostprocess):
  • inputs=['tf.Tensor(shape=(None, 56, 56, 64), dtype=float32)', 'tf.Tensor(shape=(None, 56, 56, 2), dtype=float32)', 'tf.Tensor(shape=(None, 28, 28, 64), dtype=float32)', 'tf.Tensor(shape=(None, 28, 28, 2), dtype=float32)', 'tf.Tensor(shape=(None, 14, 14, 64), dtype=float32)', 'tf.Tensor(shape=(None, 14, 14, 2), dtype=float32)']
  • training=False
  • kwargs=<class 'inspect._empty'>

I’m running Hailo Model Zoo v2.12.0 and Hailo AI SW Suite 2024-07.1

2

Hi @jrblkbelt,
You YAML file seems to follow the wrong architecture - it’s based on Yolov5. Please base it on the Yolov8
hailo_model_zoo/hailo_model_zoo/cfg/base/yolov8.yaml at master · hailo-ai/hailo_model_zoo (github.com)

3

Hi @Nadav. So I explicitly set the yaml using the following command:

hailomz compile --hw-arch hailo8l --har ./yolov8n.har --calib-path ./training/yolo/images/ --yaml ./hailo_model_zoo/hailo_model_zoo/cfg/networks/yolov8n.yaml

Where the base is yolov8.yaml:

cat ./hailo_model_zoo/hailo_model_zoo/cfg/networks/yolov8n.yaml

base:
- base/yolov8.yaml

and I still get the same error. I think I was always using the right yaml, I just assumed the absolute base one was the most important.

So for yolo8n, here’s all the files:

cat ./hailo_model_zoo/hailo_model_zoo/cfg/networks/yolov8n.yaml

base:
- base/yolov8.yaml
postprocessing:
  device_pre_post_layers:
    nms: true
  hpp: true
network:
  network_name: yolov8n
paths:
  network_path:
  - models_files/ObjectDetection/Detection-COCO/yolo/yolov8n/2023-01-30/yolov8n.onnx
  alls_script: yolov8n.alls
  url: https://hailo-model-zoo.s3.eu-west-2.amazonaws.com/ObjectDetection/Detection-COCO/yolo/yolov8n/2023-01-30/yolov8n.zip
parser:
  nodes:
  - null
  - - /model.22/cv2.0/cv2.0.2/Conv
    - /model.22/cv3.0/cv3.0.2/Conv
    - /model.22/cv2.1/cv2.1.2/Conv
    - /model.22/cv3.1/cv3.1.2/Conv
    - /model.22/cv2.2/cv2.2.2/Conv
    - /model.22/cv3.2/cv3.2.2/Conv
info:
  task: object detection
  input_shape: 448x448x3
  output_shape: 1x5x100
  operations: 8.74G
  parameters: 3.2M
  framework: pytorch
  training_data: coco train2017
  validation_data: coco val2017
  eval_metric: mAP
  full_precision_result: 37.23
  source: https://github.com/ultralytics/ultralytics
  license_url: https://github.com/ultralytics/ultralytics/blob/main/LICENSE
  license_name: GPL-3.0

cat ./hailo_model_zoo/hailo_model_zoo/cfg/base/yolov8.yaml
base:
- base/yolo.yaml
postprocessing:
  meta_arch: nanodet_v8
  anchors:
    scale_factors:
    - 0.5
    - 0.5
    regression_length: 15
  device_pre_post_layers:
    sigmoid: true
  nms_max_output_per_class: 300
  post_nms_topk: 300
  nms_iou_thresh: 0.7
  score_threshold: 0.001

and then the root yolo.yaml that I supplied in the original post.

4

And for good measure, here’s my alls and nms_config:

cat ./hailo_model_zoo/hailo_model_zoo/cfg/alls/generic/yolov8n.alls

normalization1 = normalization([0.0, 0.0, 0.0], [255.0, 255.0, 255.0])
change_output_activation(conv42, sigmoid)
change_output_activation(conv53, sigmoid)
change_output_activation(conv63, sigmoid)
nms_postprocess("../../postprocess_config/yolov8n_nms_config.json", meta_arch=yolov8, engine=cpu)

cat ./hailo_model_zoo/hailo_model_zoo/cfg/postprocess_config/yolov8n_nms
_config.json
{
        "nms_scores_th": 0.2,
        "nms_iou_th": 0.7,
        "image_dims": [
                448,
                448
        ],
        "max_proposals_per_class": 100,
        "classes": 1,
        "regression_length": 16,
        "background_removal": false,
        "bbox_decoders": [
                {
                        "name": "yolov8n/bbox_decoder41",
                        "stride": 8,
                        "reg_layer": "yolov8n/conv41",
                        "cls_layer": "yolov8n/conv42"
                },
                {
                        "name": "yolov8n/bbox_decoder52",
                        "stride": 16,
                        "reg_layer": "yolov8n/conv52",
                        "cls_layer": "yolov8n/conv53"
                },
                {
                        "name": "yolov8n/bbox_decoder62",
                        "stride": 32,
                        "reg_layer": "yolov8n/conv62",
                        "cls_layer": "yolov8n/conv63"
                }
        ]
}

I guess I did update the image_dims here in the nms too. Forgot about that.

5

And is it good now?
The error clearly suggests that something doesn’t add up between the output shapes from the NN model and the ones that the NMS add-on expects

6

I found the issue. I actually had 2 classes, not 1, so had to set that in the nms_config and pass it to the --classes 2 arg of the compile command.

But now I have a new error during Quantization.

[info] Stats Collector is done (completion time is 00:00:21.84)
[info] Starting Fix zp_comp Encoding
[info] Fix zp_comp Encoding is done (completion time is 00:00:00.00)
[info] matmul_equalization skipped
Traceback (most recent call last):
...
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/algorithms/neg_exponent_fixer/neg_exp_fixer.py", line 74, in _run_int
    l_fix = self.fix_output(l_fix)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/algorithms/neg_exponent_fixer/neg_exp_fixer.py", line 120, in fix_output
    self._log_negative_exponent_shift(l_fix)
  File "/local/workspace/hailo_virtualenv/lib/python3.8/site-packages/hailo_model_optimization/algorithms/neg_exponent_fixer/neg_exp_fixer.py", line 224, in _log_negative_exponent_shift
    raise NegativeSlopeExponentNonFixable(
hailo_model_optimization.acceleras.utils.acceleras_exceptions.NegativeSlopeExponentNonFixable: Quantization failed in layer yolov8n/conv63 due to unsupported required slope. Desired shift is 12.0, but op has only 8 data bits. This error raises when the data or weight range are not balanced. Mostly happens when using random calibration-set/weights, the calibration-set is not normalized properly or batch-normalization was not used during training.

I tried to update the images in the --calib-path ./training/yolo/images/to be all 448x448 (they were all actually 446x172 and the training didn’t complain) but still got the same error. Any insight into this error?

7

I did find this: facing issue for onnx to hec for custom trained model. - #2 by Omer

I’ll try and relabel and retrain on the images that are the appropriate 448x448 but batch normalization is definitely happening on yolo training.

8

Unfortunately I relabeled and retrained on 100% 448x448 images and still getting the same error. Could definitely use some more ideas on what I can check.

9

Check this thread:

10

That all worked! I even trained and compiled a YoloV8m model and that needed a slightly different set of conv #s but wasn’t hard to adapt the fix to it. Thanks!

1 Like